| Field | Value |
|---|---|
| Standard | Internationalization Tag Set (ITS) Version 2.0 |
| W3C Document | W3C Recommendation, 29 October 2013 |
| Latest Version | https://www.w3.org/TR/its20/ |
| Authority | World Wide Web Consortium (W3C) — MultilingualWeb-LT Working Group |
| Spec Format | XML Schema (XSD) + Relax NG (RNG) + RDF vocabulary (itsrdf: namespace) + HTML prose |
| npm Package | @amlhubs/its |
| npm Version | 0.0.1 |
| Peer Dependencies | @amlhubs/uml, @amlhubs/mof |
| License | MIT |
The Internationalization Tag Set (ITS) Version 2.0 is the W3C-Recommendation metamodel that prescribes how content authors, localization tool vendors, machine-translation engines, and translators express internationalization and localization metadata directly in the source content (XML, HTML5, or any standoff annotation format). ITS 2.0 supersedes ITS 1.0 (2007) and was published as a W3C Recommendation on 29 October 2013 by the MultilingualWeb-LT Working Group, the only W3C body chartered to specify content-internationalization metadata at the document-fragment level. ITS 2.0 governs nineteen (19) data categories — Translate, Localization Note, Terminology, Directionality, Ruby, Language Information, Elements Within Text, Domain, Disambiguation, Provenance, External Resource, Target Pointer, ID Value, Preserve Space, Localization Quality Issue, Localization Quality Rating, MT Confidence, Allowed Characters, Storage Size, Locale Filter — and binds each through three orthogonal mechanisms: local markup (per-element XML/HTML attributes in the its: namespace), global rules (XPath-selected <its:rules> blocks separated from content), and standoff RDF (the itsrdf: vocabulary that maps every data category onto RDF properties for ontological interchange).
The @amlhubs/its npm package repackages the ITS 2.0 metamodel as extensible TypeScript interfaces and base classes. The package surfaces every data category as a CMOF-compliant Class with strongly typed Property slots, every closed value-set as an Enumeration, and every cross-category relationship as a typed association — citing the precise §-section of the W3C Recommendation in the JSDoc header above each declaration so that every emitted symbol is an auditable projection of the W3C specification rather than a TypeScript invention.
Adopting ITS through a typed package produces three compounding business levers.
The first lever is regulatory and industry-standards alignment for localization-bearing content. ITS 2.0 is the W3C-Recommendation reference for translation-agnostic metadata in the localization industry — every major TMS (Translation Management System) vendor (memoQ, SDL Trados Studio, XTM Cloud, Wordfast, Smartcat, Crowdin, Lokalise, Phrase) consumes ITS-annotated XML/HTML through the XLIFF 2.x interchange format standardized by OASIS and through TBX (TermBase eXchange) for terminology. Because the interfaces @amlhubs/its exports project the W3C ITS 2.0 Recommendation directly, content emitted against them flows into every one of those tools without a custom converter. Ventures that would otherwise spend quarters rebuilding proprietary localization-metadata exporters for each new TMS contract amortize that engineering cost to zero.
The second lever is agentic runtime leverage for multilingual NLP pipelines. Ageni's Probabilistic Reduction Engine consumes the ITS metamodel as the deterministic substrate over which large-language-model translation, terminology extraction, MT post-editing, and quality estimation operate. When an agent annotates source content against ILocQualityIssue, IMtConfidence, IProvenance, or ITerminology, the TypeScript compiler evaluates whether the annotation is a well-formed ITS construct at the same moment the compiler evaluates whether the code itself is well-formed. Localization-quality regressions that would otherwise slip past a natural-language review (inventing a quality-issue type, misattributing an MT engine, violating the locQualityIssueScore 0–100 range, omitting a provenanceRecordsRef on a tool-modified segment) are caught at compile time, and every surviving annotation reference traces to a §-section of the Recommendation through the JSDoc header.
The third lever is compounding reuse across the AML metamodel stack. ITS 2.0 sits one layer above the OMG core: it depends on @amlhubs/uml for the Class/Property/Enumeration baseline that types every data category, and on @amlhubs/mof for the reflective Tag/Factory machinery that lets a runtime reflection ring inspect ITS annotations as first-class metamodel instances. Downstream NLP packages (@amlhubs/lmf-core for lexical resources, @amlhubs/maf for morpho-syntactic annotation, @amlhubs/laf-graf for offset-anchored linguistic annotation, @amlhubs/iso-24617-2 for SemAF dialogue acts) all consume ITS for their localization-aware variants. Owning the typed ITS package gives the ageni platform one coherent localization surface across every NLP venture (illummaa.com's trilingual content engine, lesvolsarabais.ca's FR/EN/ES newsletter pipeline, ageni.io's marketplace localization layer) rather than re-inventing four loosely-coupled localization vocabularies.
The package surfaces every ITS 2.0 data category and the supporting structural metaclasses (rules block, selectors, parameters, standoff vocabulary). The complete enumeration lives in its.ts; the table below summarizes the data categories and cites the authoritative §-section.
| ITS 2.0 Data Category | §Section | Surfaced Class |
|---|---|---|
| Translate | §6.2 | ITranslate |
| Localization Note | §6.3 | ILocNote |
| Terminology | §6.5 | ITerminology |
| Directionality | §6.6 | IDir |
| Ruby | §6.7 | IRuby |
| Language Information | §6.8 | ILanguageInformation |
| Elements Within Text | §6.9 | IWithinText |
| Domain | §6.10 | IDomain |
| Disambiguation | §6.11 | IDisambiguation |
| Provenance | §6.12 | IProvenance |
| External Resource | §6.13 | IExternalResource |
| Target Pointer | §6.14 | ITargetPointer |
| ID Value | §6.15 | IIdValue |
| Preserve Space | §6.16 | IPreserveSpace |
| Localization Quality Issue | §6.17 | ILocQualityIssue |
| Localization Quality Rating | §6.18 | ILocQualityRating |
| MT Confidence | §6.19 | IMtConfidence |
| Allowed Characters | §6.20 | IAllowedCharacters |
| Storage Size | §6.21 | IStorageSize |
| Locale Filter | §6.4 | ILocaleFilter |
Every interface is accompanied by an extensible base class with the same name minus the I prefix (e.g., Translate, LocNote, LocQualityIssue). The full list and the JSDoc headers citing each §-section live at its.ts.
@amlhubs/its sits one layer above the OMG core. It depends on UML for the metaclass spine and on MOF for the reflective ring.
@amlhubs/uml (root, zero dependencies)
▲
│ peerDependency
├── @amlhubs/mof (reflective machinery over UML)
│ ▲
│ │ peerDependency
│ └── @amlhubs/its (this package — ITS 2.0 data categories typed against UML+MOF)
│
└── (other downstream metamodels — ocl, sbvr, smm, vdml, xmi)
npm install @amlhubs/itsimport type {
ITranslate,
ILocQualityIssue,
IProvenance,
IRulesElement,
} from '@amlhubs/its';
// Mark a content fragment as not translatable.
const noTranslate: ITranslate = {
elementId: 'ITS_Translate_disclaimer',
translate: 'no',
};
// Record a localization-quality issue in machine-readable form.
const issue: ILocQualityIssue = {
elementId: 'ITS_LocQualityIssue_term_drift',
locQualityIssueType: 'terminology',
locQualityIssueComment: 'Inconsistent rendering of "carbon offset" in fr-CA segments.',
locQualityIssueSeverity: 75,
locQualityIssueProfileRef: 'http://example.org/qualityProfiles/quebec-french',
};The source artifact is its.ts. Every interface JSDoc header declares @standard W3C ITS 2.0 -- Recommendation 29 October 2013 and a @section §x.y reference.
- W3C ITS 2.0 Recommendation (29 October 2013)
- W3C MultilingualWeb-LT Working Group
- W3C Internationalization Activity
- ITS 2.0 schemas (XSD/RNG)
- ITS 2.0 RDF vocabulary (
itsrdf:) - ITS 2.0 test suite
- Companion: XLIFF 2.x — OASIS
- Companion: TBX — TermBase eXchange (LISA)
| Version | Date | Change Summary |
|---|---|---|
| 0.0.1 | 2026-05-03 | Initial publish — all 19 ITS 2.0 data categories surfaced as CMOF-typed Classes with Property slots and Enumeration value sets; supporting IRulesElement, selectors, and standoff itsrdf: vocabulary included. |
MIT — see LICENSE.